In order to ensure a high yield in production of semiconductors, it is important to take measures by finding defects generated in the production process at an early stage.
A SEM (Scanning Electron Microscope) defect observation apparatus is an apparatus for observing a variety of such defects, and is typically an apparatus for observing an image of defect coordinates, which has been detected with a higher-level defect inspection apparatus, with higher image quality than that of the higher-level defect inspection apparatus. Specifically, a sample stage is moved to the defect coordinates output from the higher-level defect inspection apparatus, and then, an image is captured at a low magnification at about a level that allows the field of view to contain the target defect to be observed. Then, the defect coordinates are detected from the captured low-magnification image, and the sample stage is moved so that the defect is located at the center of the field of view or the center of image capturing is moved, and then, a high magnification image for observation is acquired at a high magnification that is suitable for observing, the defect. As described above, defect coordinates are detected from a low-magnification image since defect coordinates output from a higher-level defect inspection apparatus contain errors within the range of the specifications of the apparatus. Thus, when a high-quality defect image is acquired with a SEM defect observation apparatus, a process of correcting such errors should be performed.
ADR (Automatic Defect Review or Redetection) is a technique of automating a step of acquiring such a high-quality defect image. In the ADR, it is necessary to optimize the conditions to acquire a low-magnification image for detecting a defect, the conditions to acquire a high-magnification image for analyzing the defect, and the like in accordance with the defect detection accuracy of the higher-level defect inspection apparatus, the properties of a sample, and the type of the defect to be observed in order to achieve both a high detection rate of the ADR and high throughput.
ADC (Automatic Defect Classification) is a technique of automating a step of classifying defects by type on the basis of a defect image acquired with high quality. As the ADR has been put into practical use and the applicable step of the ADR has increased, there have been cases where the operation of analyzing the automatically acquired defect image is a bottleneck, and the applicable step of the ADC is thus desired to be increased. In practice, the ADC has been put into practical use, and the applicable step of the ADC has increased, in particular, in the mass production line.
Patent Literature 1 discloses a technique of executing an autofocus process outside the field of view of a defect image so that contamination generated in the autofocus process will not adversely affect the defect image.